SCIENTIFIC ARTICLE

Spondylodiscitis following endovascular abdominal aortic aneurysmrepair: imaging perspectives from a single centre’s experience

Ramin Mandegaran1& Christopher S. W. Tang1

& Erlick A. C. Pereira2 & Ali Zavareh1

Received: 13 January 2018 /Revised: 19 March 2018 /Accepted: 22 March 2018 /Published online: 14 April 2018# The Author(s) 2018

AbstractObjective Very few reports have previously described spondylodiscitis as a potential complication of endovascular aortic aneu-rysm repair (EVAR). We present to our knowledge the first case series of spondylodiscitis following EVAR based on ourinstitution’s experience over an 11-year period. Particular attention is paid to the key imaging features and challenges encounteredwhen performing spinal imaging in this complex patient group.Materials and Methods Of 1,847 patients who underwent EVAR at our institution between January 2006 and January 2017, atotal of 9 patients were identified with imaging features of spondylodiscitis (0.5%). All cross-sectional studies before and afterEVAR were assessed by a Consultant Musculoskeletal Radiologist and a Musculoskeletal Radiology Fellow to evaluate forfeatures of spondylodiscitis.Results All 9 patients had single-level spondylodiscitis involving lumbosacral levels adjacent to the aortic/iliac stent graft. Eightout of nine patients had an extensive anterior paravertebral phlegmon/abscess that was contiguous with the infected stent graftand native aneurysm sac ± anterior vertebral body erosion. Epidural disease was present in only 3 out of 9 patients and was aminor feature. MRI was non-diagnostic in 3 out of 9 patients owing to susceptibility artefact. 18F-FDG PET/CT accuratelydepicted the spinal level involved and adjacent paravertebral disease in patients with non-diagnostic MRI and was adopted as thefollow-up modality in 3 out of 5 surviving patients.Conclusion Spondylodiscitis is a rare complication post-EVAR. Imaging features of disproportionate anterior paravertebraldisease and anterior vertebral body bony involvement suggest direct spread of infection posteriorly to the adjacent vertebralcolumn. Use of MRI versus 18F-FDG PET/CT as the optimal imaging modality should be directed by the type of stent graftdeployed.

Keywords Spondylodiscitis . EVAR .MRI . PET/CT

Introduction

Since it was first described in 1991 [1], endovascular aorticaneurysm repair (EVAR) has established itself over open sur-gical repair as the preferred method for treating abdominal

aortic aneurysms in patients with suitable anatomy, given itslower mortality rate and comparable long-term survival [2, 3].Whilst aortic graft infection as a complication of EVAR hasbeen well documented in recent years, numerous studies haveshown it to be a relatively rare problem, with an incidence of0.5–1% [4–8].

Spondylodiscitis as a potential complication of EVAR hasreceived very little attention in the literature and to our knowl-edge, only 4 case reports have been published to date [9–12].We present a case series of 9 patients who went on to developspondylodiscitis following EVAR, paying particular attentionto: imaging findings and trends identified in the imaging fea-tures and discussing the challenges and optimal imaging strat-egy in this complex patient group, which is highly influencedby the type of stent graft deployed.

* Ramin Mandegaranramin1986@hotmail.com

1 Department of Radiology, Guy’s and St Thomas’ NHS FoundationTrust, 2nd Floor Tower Wing, Great Maze Pond, London SE1 9RT,UK

2 Academic Neurosurgery Unit St George’s, University of London,London SW17 0QT, UK

Skeletal Radiology (2018) 47:1357–1369https://doi.org/10.1007/s00256-018-2939-z

Materials and methods

A retrospective analysis of all patients who underwent EVARat a single tertiary centre over an 11-year period betweenJanuary 2006 and January 2017 was performed to identifypatients with features of spondylodiscitis on cross-sectionalimaging studies after EVAR. A total of 1,442 abdominalEVARs and 405 thoracic EVARs (1,847 EVARs in total) wereperformed over the 11-year period. A total of 9 patients (0.5%)were identified with imaging features of spondylodiscitis oncross-sectional imaging (comprising CT, MRI and 18F-fluorodeoxyglucose [FDG] PET-CT studies) after EVAR.For each patient, all cross-sectional studies available, bothbefore and after EVAR, were reviewed by a ConsultantMusculoskeletal Radiologist and a MusculoskeletalRadiology Fellow to evaluate for the key features ofspondylodiscitis (the presence of endplate erosion, disc en-hancement, disc abscess) and the extent of paraspinal disease(the presence of an epidural or paravertebral phlegmon/abscess and associated psoas involvement). Features consid-ered suggestive of a direct spread of infection to the spinefollowing EVAR comprised involvement of spinal levels im-mediately adjacent to the stented aorto-iliac vessels; anteriorparavertebral disease contiguous with the stent graft/nativeaneurysm sac; extensive erosion of the anterior vertebral bodymargin. Progression/resolution of imaging findings was alsoassessed on interval studies where available.

Results

Clinical features and diagnosis

All 9 patients were male, with a mean age of 73 years (range58–87 years). The main clinical features for all 9 patients, areoutlined in Table 1.

Eight patients had infra-renal EVAR for an infra-renalaortic/iliac artery pathological condition (7 patients withinfra-renal aortic aneurysms, 1 patient with an infra-renal aor-tic penetrating ulcer and 1 patient with right internal iliacartery aneurysm). One patient with an infra-renal saccular an-eurysm had a positive blood culture before EVAR, supportingthe diagnosis of a probable infective aneurysm (patient 5).Another patient had a supra-renal aortic aneurysm requiringsupra-renal extension of the EVAR (patient 9).

Clinical presentation following EVAR that led to imagingand identification of spondylodiscitis was variable. Back painwas the presenting feature in only 3 out of 9 patients. A further3 out of 9 patients presented with non-specific features ofsepsis (fever, rigours ± nausea and vomiting) and no signifi-cant pain. Of the remaining 3 patients, 1 complained of centralabdominal pain and vomiting, whereas 2 patients werecompletely asymptomatic with vertebral endplate erosion

noted incidentally on routine post-EVAR surveillance CT.Interval from initial aortic intervention to the first identifica-tion of spondylodiscitis features on imaging was also marked-ly variable, ranging from 1 to 19 months (median 3 months).

Positive microbiological cultures were identified in 6 out of9 patients, all of which were bacterial (Escherichia coli in 3patients, Staphylococcus aureus in 2 patients, Streptococcuspneumoniae in 1 patient). Two patients had positive culturesobtained from paravertebral collection biopsies and 2 patientsfrom adjacent psoas abscess biopsy. Four patients had positiveblood cultures (2 of whom additionally had positive culturesfrom paravertebral collection/psoas abscess biopsy). Spinalbiopsy of the disc level involved was performed in 2 patients,both of whom were culture-negative. No causative organismwas identified in 3 out of 9 patients.

Imaging findings

The key imaging features of spondylodiscitis identified in all 9patients are outlined in Table 2. In all 9 patients, only onesingle level lumbosacral disc involvement was demonstrated.L2/3 and L5/S1 were the most superior and inferior levelsinvolved, with no involvement of spinal levels that were notimmediately adjacent to the stent graft.

In 5 out of 9 patients, spondylodiscitis features were firstidentified on CT, demonstrating endplate erosion ± anteriorvertebral body erosion, loss of disc height and paravertebralsoft tissue/abscess (Fig. 1). All patients had at least one MRIspine (non-diagnostic in 2 patients and difficult to interpret ina third patient owing to extensive susceptibility artefact arisingfrom the adjacent stent graft) and 4 patients underwent at leastone 18F-FDG PET/CT (2 of whom had extensive artefact onMRI). All patients showed evidence of vertebral endplate ero-sion and either intervertebral disc enhancement on MRI(Fig. 2) or markedly increased tracer uptake on 18F-FDGPET/CT (Fig. 3), although only 5 out of 9 had imaging evi-dence of a disc abscess.

Eight out of nine patients had paravertebral abscesses orphlegmon of varying sizes, which were either contiguous withor immediately adjacent to the native aneurysmal aortic sac.Five of the eight patients with paravertebral abscess formationalso showed evidence of psoas abscess involvement. Only 3out of 9 patients had an epidural component, of whom 2 hadvery minor non-compressive epidural phlegmon. The thirdpatient with epidural disease had spondylodiscitis centred atL4/5 with an epidural phlegmon impinging on the traversingL5 nerve roots and exiting L4 nerve roots bilaterally in thelateral recesses and neural exit foramina respectively (Fig. 4).

Treatment, progress and outcome

All patients received antibiotic treatment according to localmicrobiological advice. Four out of nine patients died within

1358 Skeletal Radiol (2018) 47:1357–1369

Table1

Sum

maryof

clinical

details,includinginitial

aortic

pathological

condition,levelof

aortic

stentin

g,initial

presentatio

n,tim

einterval

leadingto

identificationof

spondy

lodiscitis,

and

microbiologicalcultu

reresults

Patient

Initialaorticpathologicalcondition

Locationof

stent

Presentationleadingto

identification

ofspondylodiscitis

Intervalfrom

stentin

gto

firstimagingfeatures

ofspondylodiscitis(m

onths)

Microbiologicalcultu

re(sourceof

cultu

resample)

1Infra-renalA

AA

Infra-renalaortato

bilateralC

IAs

Centralabdominalpain,v

omiting

19Culture-negative(spinal

biopsy,blood

cultu

re)

2Penetratin

ginfra-renalu

lcer

Infra-renalaortastentedinitially—displaced

inferiorly.

Re-stentedinfra-renalaortato

bilateralC

IAs

Backpain

6S.aureus

(adjacentp

soas

collection)

3Infra-renalA

AA

Infra-renalaortato

bilateralC

IAs

Vom

iting,feversandrigours.

Bonyvertebralerosion

notedatsurgery

during

removalof

infected

stentg

raft.

2E.coli(bloodcultu

re)

4Infra-renalA

AA

Infra-renalaortato

bilateralC

IAsinitially

stented.

Aortic

stentcollapsed—newinfra-renalaortic

stent

deployed

with

inpre-existin

gstentafter

6weeks

Low

erback

andrightleg

pain

3Culture-negative(spinal

biopsy,blood

cultu

re)

5Infra-renalm

ycoticAAA(E.coli

presentinpre-EVARbloodcul-

ture)

Infra-renalaortato

bilateralC

IAs

Asymptom

atic.V

ertebralendplateerosion

onsurveillanceCT

2E.coli(paravertebral

collection)

6Right

CIA

aneurysm

Infra-renalaortato

bilateralC

IAs

Asymptom

atic.V

ertebralendplateerosion

onsurveillanceCT

1S.pneumoniae

(blood

cultu

re)

7Infra-renalA

AA

Infra-renalaortato

bilateralC

IAs

Fever

4S.aureus

(paravertebral

collection)

8Right

IIAaneurysm

Infra-renalaortato

bilateralC

IAsandrightE

IA(right

IIAexcluded

bystent)

Low

erback

pain

6E.coli(bloodcultu

reandadjacent

psoas

collection)

9Suprarenalright-sided

saccular

AAA

Supra-renalabdom

inalaortato

bilateralC

IAswith

fenestratio

nsforSMAandrenalarteries

Fever

1Culture-negative

(paravertebralcollection,

bloodcultu

re)

AAAabdominalaorticaneurysm

,EVA

Rendovascular

aorticaneurysm

repair,

CIA

common

iliac

artery,IIA

internaliliac

artery,E

IAexternaliliac

artery,SMAsuperior

mesentericartery

Skeletal Radiol (2018) 47:1357–1369 1359

Table2

Summaryof

spondylodiscitisim

agingfindings,progressandoutcom

es

Patient

Imaging

modalities

demonstratin

gspondylodiscitis

features

Imagingfeatures

Progress

Outcome

Spondylodiscitis

Paraspinalphlegm

on/abscess

Spinal

level(s)

involved

End

plate

erosion

Anterior

vertebral

body

erosion

Disc

enhancem

ent/

uptake

Disc

abscessParavertebral

PsoasEpidural

1CT,

MRI

L2/3

++

++

++

+Stentrem

oval+axillobifem

oralbypass.

Progressive

increase

inparavertebral/d

iscalab-

scessover

5months

Died7monthsfollo

wingidentification

ofspondylodiscitis

2PET/CT,

MRI

L3/4

++

++

++

–Leftpsoascollectionaspirated—

partialreductio

nin

volume

Died2monthsfollo

wingidentification

ofspondylodiscitis(under

palliative

care

forrecurrentp

haryngeal

squamouscellcarcinom

a)

3MRI,CT

L4/5

++

++

++

–Stentrem

oval+axillobifem

oralbypass

Died3monthsfollo

wingidentification

ofspondylodiscitis

4MRI

L5/S1

+–

++

––

+Resolutionof

disc

abscessandepiduralphlegm

onon

serialMRI

Discharged

5CT,

MRI,

PET/CT

L4/5

++

+–

+–

+Stentg

raftremovalwith

complex

aortic

reconstructio

n.Paravertebralcollectiondrainedto

resolutio

n.L4/5debridem

ent,decompression

andfusion—

serialsubsequent

PET/CTshow

sno

significant

uptake

atthefusion

site

Discharged

6CT,

MRI,

PET/CT

L4/5

+–

+–

+–

–Not

follo

wed

upwith

imaging,crosssectional

imaging

Discharged

7CT,

MRI

L2/3

++

++

++

–Attempted

drainage

oftheparavertebralcollection

Died1month

follo

wingidentification

ofspondylodiscitis

8CT,

MRI

L5/S1

++

+–

++

–Attempted

drainage

oftherightp

soas

collection—

minim

alvolumeaspirated

Dischargedwith

follo

w-upPET/CT

surveillance—

didnotattend

9CT,

MRI,

PET/CT

L1/2

++

+–

+–

–Reductio

ninthesize

oftheparavertebralcollection

anduptake

relatedtocollectionanddiscspaceon

intervalPET/CT

Dischargedwith

follo

w-upPET/CT

surveillance

1360 Skeletal Radiol (2018) 47:1357–1369

7 months of first imaging identification of spondylodiscitis (3patients secondary to vascular complications; 1 patient withadvanced recurrent head and neck malignancy), of which 2underwent stent graft removal and axillobifemoral bypassgraft.

Of the 5 out of 9 patients who survived, only oneunderwent removal of the stent graft, with aortic reconstruc-tion, vertebral debridement, decompression and L4/L5 spinalfusion. The remaining 4 patients were successfully treatedconservatively ± attempted percutaneous drainage.

18F-FDG PET/CT was adopted as the follow-up imagingmodality in 3 patients owing to extensive susceptibility arte-fact on MRI. Two of these patients demonstrated either reduc-tion or resolution of increased uptake at the involved disc andadjacent paravertebral disease, one of whom was treated con-servatively (Fig. 5), the other undergoing debridement, de-compression and L4/5 fusion (patient 5). The third patientdid not attend the follow-up imaging.

One patient was followed up with serial MRI studies show-ing the resolution of disc abscess and epidural phlegmon (pa-tient 4, Fig. 6). The final surviving patient did not undergo anyinterval MRI or 18F-FDG PET/CT following initial diagnosisof spondylodiscitis (patient 6).

Discussion

Infection of the stent graft following EVAR is a potentiallydevastating but uncommon complication, with an incidence ofbetween 0.5 and 1% and a mortality rate of 11% and 30% at 1

and 12 months in treated patients [4–8]. Although these stud-ies have shown the tendency of infected stent grafts to result inthe spread of infection to the native surrounding aortic sac andpara-aortic soft tissues, it is perhaps surprising that only fourprevious case reports have described spondylodiscitis as acomplication of EVAR given the close proximity of the ver-tebral column and aorta [9–12]. In all of these studies, CTimaging demonstrated features of an infected stent graft andnative aneurysm sac with abscess formation contiguous with asingle adjacent lumbar level with features of spondylodiscitis[9–12].

Several studies have suggested predisposing risk factorsthat may contribute to infection of the stent graft followingEVAR, including multiple endovascular/surgical procedures,an immunocompromised state and nosocomial blood streamsepticaemia [13–15]. Although specific predisposing risk fac-tors for spondylodiscitis following EVAR are difficult to iden-tify given our relatively small cohort, we found that all 9patients had single-level spondylodiscitis at lumbosacrallevels adjacent to the stented aorto-iliac vessels.Furthermore, 8 of the 9 patients had spondylodiscitis withadjacent paravertebral collections that were either contiguouswith or immediately adjacent to the native aneurysm, 7 ofwhom had not only endplate, but also anterior vertebral bodyerosion (Fig. 1). Although the small cohort size precludes anydefinitive conclusions, the pattern of extensive predominantlyanterior vertebral bony and paravertebral disease suggests di-rect spread from the adjacent infected aortic stent graft andnative aneurysm sac as a likely route of vertebral columninfection in most patients in our cohort. However,

Fig. 1 An 87-year-old man (pa-tient 1). a Axial and b coronal CT19 months followingendovascular aortic aneurysmrepair (EVAR), demonstrating L3anterolateral vertebral body andsuperior endplate erosion.Peripherally enhancingparavertebral/left psoas abscess iscontiguous with the markedlyexpanded native aneurysmencasing the stent graft. c AxialCT after a further 12 months withinterval stent graft removal andaxillobifemoral bypass. There isnow more progressive vertebralbody destruction with acontiguous abscess nowinvolving the native excludedaneurysm sac, paravertebral softtissues and replacing both psoasmuscles

Skeletal Radiol (2018) 47:1357–1369 1361

haematogenous spread of infection could also present withextensive anterior paravertebral disease. In only 1 patient (pa-tient 4) was the paravertebral component not associated withthe native aneurysm sac, suggesting haematogenous seedingof infection to the involved disc rather than direct spread inthis particular case (Fig. 6). These findings highlight the pointthat patients with infected stent grafts and extensive para-aortic infection are potentially at a higher risk of adjacent levelspondylodiscitis and anterior vertebral body osteomyelitis.This may also account for why relatively few patients in ourcohort (3 out of 9 patients) had disease extension posterior tothe epidural space (Fig. 4).

Only one patient in our cohort had positive pre-EVARblood culture in addition to typical saccular aneurysmal mor-phology, consistent with infective aneurysm (Fig. 7). The or-ganisms identified on pre-EVAR blood cultures and post-EVAR paravertebral collection biopsies were similar(E. coli). Presence of an infective aneurysm and pre-existing

septicaemia before any vascular intervention has been shownto be associated with the development of spondylodiscitis[16–20].

It is notable that although all 9 patients in our study dem-onstrated imaging features of spondylodiscitis, a positive cul-ture was isolated in only 6 patients from a combination ofblood cultures and/or CT-guided paravertebral/psoas collec-tion biopsies. The 3 patients with no causative organism iden-tified comprised 2 patients with culture-negative spinal biop-sies and 1 patient with a culture-negative biopsy of theparavertebral collection. This is in line with previous studiesthat have highlighted the limited positive yield of CT-guidedspinal biopsies (42–57%), and a moderately higher yield ob-tained from the biopsy of a paravertebral phlegmon/abscess(68%) [21–23].

With regard to the imaging assessment of spondylodiscitisfollowing EVAR, our retrospective cohort of patientsunderwent varying combinations of CT, MRI and/or 18F-

Fig. 2 A70-year-oldman (patient3). a Sagittal T2, b T1 pre-contrast and c T1 post-contrastsequences demonstratingincreased T2-weighted signalcentred on the L4/5 disc withanterior paravertebral extensioncorresponding to a peripherallyenhancing abscess. d Axial T1pre-contrast and e T1 post-intravenous contrast medium atthe L4/5 disc level showcontiguous peripherallyenhancing disc/anteriorparavertebral abscess. Note thelack of extension posterior to theepidural space

1362 Skeletal Radiol (2018) 47:1357–1369

FDG PET/CT, highlighting the dilemma faced when imagingthis challenging patient group. Across these modalities, thetypical features of a bacterial spondylodiscitis were identifiedin all 9 patients—namely single-level loss of disc height,erosion/destruction of adjacent vertebral endplate cortices,signal abnormality and enhancement/tracer uptake centredon the disc involved and adjacent endplates ± paravertebral/epidural enhancing phlegmon/abscess [24]. As mentionedpreviously, the disproportionate extent of anteriorparavertebral disease in most patients and associated anteriorvertebral body erosion are suggestive of the direct spread ofinfection to the spine.

In most cases, contrast-enhanced CT was the first cross-sectional modality performed owing to the non-specific pre-senting features of back/abdominal pain, fever, nausea andvomiting. More concerning was the detection ofspondylodiscitis on routine post-EVAR surveillance CT in 2

patients who were asymptomatic, highlighting the often insid-ious course of spondylodiscitis. In 5 out of 9 patients,spondylodiscitis features were first identified on CT, demon-strating established endplate destruction ± anterior vertebralerosion, loss of disc height and paravertebral soft tissue(Figs. 1, 7). CTwas also able to depict morphological changesin the native aneurysm sac compared with preoperative stud-ies, in addition to new aortic rim enhancement and internal gaslocules, indicating infection of the stent graft and aneurysmsac (Fig. 8). However, assessment of epidural disease anddifferentiating between paravertebral phlegmon and abscesswas challenging on CT.

Gadolinium-enhancedMRI is considered the gold standardin the assessment of spondylodiscitis, owing to its superiorsoft-tissue contrast permitting accurate assessment ofparavertebral and epidural extension. The contrast enhance-ment is also superior in differentiating between a phlegmon

Fig. 3 A75-year-oldman (patient2). a Sagittal T2 and b STIRsequences demonstrating markedsusceptibility artefact limitingimage interpretation. Increasedsignal can be seen in the L3/4 discin a, but no assessment of theparavertebral soft tissues can bemade. c Sagittal and d axial PET/CT demonstrate abnormalincreased uptake related to theEVAR stent graft with soft-tissuedensity and increased uptakeextending into the L3/4 discspace, where there is endplateerosion and loss of disc heightconsistent with spondylodiscitis.Expansion of the left psoas withperipheral uptake suggests asso-ciated psoas abscess formation.Diffusely increased marrowuptake elsewhere in the spine wasthought to be reactive

Skeletal Radiol (2018) 47:1357–1369 1363

and an abscess, in addition to the extent of marrow infiltration[24]. However, in the context of the post-EVAR patient, theadded complication of susceptibility artefact from the adjacentmetallic aorto-iliac stents must be considered. This may resultin varying degrees of in-plane distortion (signal loss and pile-up), through-plane artefact resulting in geometric distortion,and ineffective fat suppression [25, 26]. In our cohort, theextent of susceptibility artefact was widely variable, rangingfrom minimal image distortion to non-diagnostic images,

which precluded accurate assessment of the spine in 3 out of9 patients. This is in large part due to the specific stent typedeployed, given that currently used materials in constructionof the stent skeleton may comprise nitinol (nickel titaniumalloy), cobalt chromium or stainless steel alloys [12].Nitinol-based stent grafts have the lowest magnetic suscepti-bility and generate minimal artefact, whereas the greatest de-gree of artefact is observed with stainless steel grafts, particu-larly those of 304L stainless steel alloys, owing to their

Fig. 4 A58-year-oldman (patient5). a Sagittal T1, b T1 post-contrast and c axial T1 post-contrast 3 months followingEVAR. There is an L4/5 discabscess with a paravertebral andepidural enhancing phlegmonextending into the lateral recessesbilaterally

1364 Skeletal Radiol (2018) 47:1357–1369

ferromagnetic properties [27, 28]. However, the extent of ar-tefact may be amplified or minimised significantly dependingon the type of sequence and specific sequence parametersutilised. Basic methods to reduce such artefacts include theuse of lower field strength, the use of spin-echo/turbo spin-echo sequences with long echo trains rather than gradientecho, and the use of short inversion time recovery (STIR)sequences for fat suppression rather than conventional spectralfat suppression [25, 26, 29]. These conventional sequencescan be further optimised to reduce metallic artefacts by in-creasing receiver bandwidth, using a larger matrix to increase

in-plane resolution and thinner slices to increase through-plane resolution (albeit at a cost of reduced overall signal-to-noise ratio (SNR)—this may be boosted with increased signalaverages) [25, 26, 29]. Figure 6 shows how the application ofsome of these methods of optimisation allowed serial MRIs tobe performed in patient 4 in the context of a 316L stainlesssteel stent (Advanta V12; Atrium Medical Corporation) withminimal stent-related susceptibility artefact, clearly depictingthe interval resolution of spondylodiscitis. Of note, this wasthe only patient in whom MRI was adopted as the follow-upimaging modality and accurately demonstrated interval

Fig. 5 A82-year-oldman (patient9). Serial PET/CT studies. aSagittal and b axial PET/CTstudyimages performed 2 monthsfollowing EVAR demonstratingintense FDG uptake at level of theL1/2 disc, extending into theadjacent vertebral bodies. Notethe intense anterior vertebral bodyuptake at L2 with further uptakeextending into the paravertebralsoft tissues, continuous with thedilated native aneurysm sac. cSagittal and d axial images of thePET/CT study performed at a 12-month interval at similar levelsfollowing conservativemanagement, demonstratingsignificantly reduced vertebraland paravertebral uptake. Thepreviously demonstratedparavertebral abscess componentcontiguous with the nativeaneurysm sac had also resolved

Skeletal Radiol (2018) 47:1357–1369 1365

disease reduction. Figure 9 shows the vast difference in theextent of susceptibility artefact when compared with patient 8in the context of a 304L stainless steel stent (Zenith Spiral;Cook Medical) with similar imaging parameters.

More recently, various three-dimensional multispectral im-aging (3D-MSI) sequences have been developed, includingmulti-acquisition with variable-resonance image combination(MAVRIC) and slice encoding for metal artefact correction(SEMAC), which, in the context of spinal MRI in patientswith metallic spinal fixation, have been shown to overcomethe susceptibility of stainless steel or cobalt chromium alloyswith an up to eight-fold reduction in the scan volume affectedby artefact compared with conventional sequence optimisa-tion for metal artefact reduction [25, 30–32]. The 3D-MSIsequences were not used in our patient cohort, mainly becauseof their retrospective nature. However, their potential inminimising susceptibility artefact and improving image qual-ity in the context of post-EVAR spinal imaging should beexplored.

In 2 patients with significant artefact on MRI precludingaccurate assessment, 18F-FDG PET/CT proved to be of sig-nificant diagnostic value, as it accurately depicted markedlyincreased uptake at the spinal level involved in addition to thesurrounding paravertebral disease, which could not bevisualised on MRI (Fig. 3). Furthermore, 18F-FDG PET/CTwas selected as the follow-up imaging modality in 3 patientsin whom MRI had been markedly degraded by metallic

susceptibility artefact from the stent graft in 2 patients, and athird patient who had undergone spinal debridement of thevertebral column involved and spinal fixation, precluding sub-sequent accurateMRI assessment of residual spondylodiscitis.The 2 patients who underwent follow-up 18F-FDG PET/CTdemonstrated either reduction or resolution of spinal/paravertebral tracer uptake and a reduction in size of theparavertebral collection on interval studies (Fig. 5). Theseexamples highlight the potential role of 18F-FDG PET/CT,not just for diagnosis, but also as an effective means of mon-itoring the progress of spinal infection in the context of previ-ous EVAR, or indeed metallic spinal instrumentation that maypreclude accurate assessment by MRI. This is supported byseveral studies that have shown 18F-FDG PET/CT to be atleast equivalent to MRI in diagnosing spondylodiscitis andmay be used as a means of monitoring the response to treat-ment [33–37]. In the absence of PET/CT facilities, other nu-clear medicine modalities including 99mTc-MDP bone scanand Gallium scan (particularly when combined with singlephoton emission computed tomography (SPECT)/CT) havealso been shown to be viable alternatives for the detection ofspondylodiscitis [38, 39].

Our study had a few limitations: the described imagingtrends may not be definitive owing to the small sample size.The retrospective nature was another limitation, resulting indifferingMR scanning protocols and imaging parameters withvariable use of metal artefact reduction sequences.

Fig. 6 A 59-year-old man (patient 4). Sagittal T2 images from serial MRI studies demonstrating gradual resolution of an L5/S1 disc abscess extendinginto the adjacent endplates. a 3 months following EVAR, b 6-week interval following the initial MRI, c 3-month interval and d 14-month interval

1366 Skeletal Radiol (2018) 47:1357–1369

Fig. 7 A58-year-oldman (patient5). a Sagittal and b axial CTangiogram pre-EVAR. There issaccular aneurysmal dilatation atthe L3/4 level with adjacent gaslocules typical of an infectiveaneurysm. E. coli bacteraemiaconfirmed the infective aneurysm.c Sagittal and d axial images fromroutine surveillance CT 2 monthspost-EVAR demonstrate new L4/5 endplate erosion, anterior L4vertebral body erosion, gaslocules in the disc space andparavertebral soft-tissue densityextending to the stent graft

Fig. 8 A70-year-oldman (patient3). a Axial and b sagittal CTimages 2 months post-EVAR.Note gas locules within theperipherally enhancing nativeaneurysm sac extendingposteriorly to the vertebralcolumn. There is a gas-containingright psoas/paravertebral abscesscontiguous with the nativeaneurysm sac

Skeletal Radiol (2018) 47:1357–1369 1367

Conclusion

To the best of our knowledge, we have reported the first caseseries of spondylodiscitis post-EVAR—a rare complication typ-ically occurring in the context of an infected stent graft withextensive para-aortic disease. Clinicians and radiologists alikeshould be alert to the possibility of spondylodiscitis as a com-plication of EVAR given the often non-specific and insidiousclinical presentation. The imaging features of disproportionateanterior paravertebral disease and anterior vertebral body bonyinvolvement suggest direct spread of infection to the adjacentvertebral column. The optimal imaging strategy for diagnosisand follow-up is challenging and should be directed by the typeof stent graft deployed, as certain stainless steel alloys result insignificant artefact that precludes accurate assessment by MRI.In such cases, 18F-FDG PET/CT is an accurate alternative thatmay be used for early diagnosis and monitoring the diseaseresponse to treatment. Experimenting with new MR sequencessuch as 3D-MSI should be considered.

Acknowledgements The authors thank the PET department, Guy’s andSt Thomas’ NHS Foundation Trust, for the use of PET/CT images.

Compliance with ethical standards

Ethical approval All procedures performed in studies involving humanparticipants were in accordance with the ethical standards of the institu-tional and/or national research committee and with the 1964 Declarationof Helsinki and its later amendments or comparable ethical standards.

Conflicts of interest The authors declare that they have no conflicts ofinterest.

Open Access This article is distributed under the terms of the CreativeCommons At t r ibut ion 4 .0 In te rna t ional License (h t tp : / /creativecommons.org/licenses/by/4.0/), which permits unrestricted use,distribution, and reproduction in any medium, provided you give appro-priate credit to the original author(s) and the source, provide a link to theCreative Commons license, and indicate if changes were made.

References

1. Parodi JC, Palmaz JC, Barone HD. Transfemoral intraluminal graftimplantation for abdominal aortic aneurysms. Ann Vasc Surg.1991;5:491–9.

2. Greenhalgh RM, Brown LC, Kwong GP, Powel JT, Thompson SG,EVAR trial participants. Comparison of endovascular aneurysmrepair with open repair in patients with abdominal aortic aneurysm(EVAR trial 1), 30-day operative mortality results: randomised con-trolled trial. Lancet. 2004;364:843–8.

3. The United Kingdom EVAR Trial Investigators, Greenhalgh RM,Brown LC, et al. Endovascular versus open repair of abdominalaortic aneurysm. N Engl J Med. 2010;362:1863–71.

4. Smeds MR, Duncan AA, Harlander-Locke MP, et al. Treatment andoutcomes of aortic endograft infection. J Vasc Surg. 2016;63:332–40.

5. Cernohorsky P, Reijnen MM, Tielliu IF, van Sterkenburg SM, vanden Dungen JJ, Zeebregts CJ. The relevance of aortic endograftprosthetic infection. J Vasc Surg. 2011;54:327–33.

6. Ducasse E, Calisti A, Speziale F, et al. Aortoiliac stent graft infec-tion: current problems and management. Ann Vasc Surg. 2004;18:521–6.

7. Fatima J, Duncan AA, de Grandis E, et al. Treatment strategies andoutcomes in patients with infected aortic endografts. J Vasc Surg.2013;58:371–9.

8. Laser A, Baker N, Rectenwald J, Eliason JL, Criado-Pallares E,Upchurch GR Jr. Graft infection after endovascular abdominal aor-tic aneurysm repair. J Vasc Surg. 2011;54:58–63.

Fig. 9 a Sagittal turbo spin echo(TSE) T2 in patient 4 (a 59-year-old man) with a 316L stainlesssteel Advanta V12 (AtriumMedical Corporation) stent insitu. Note the minimalsusceptibility artefact comparedwith b sagittal TSE T2 in patient 8(an 84-year-old man) with a 304Lstainless steel Zenith Spiral (CookMedical) stent in situ. Bothsequences were obtained at 1.5 Twith similar imaging parameters,differing only in the number ofsignal averages (3 in a, 2 in b)

1368 Skeletal Radiol (2018) 47:1357–1369

9. De Koning HD, van Sterkenburg SM, Pierie ME, Reijnen MM.Endovascular abdominal aortic aneurysm repair complicated byspondylodiscitis and iliaco-enteral fistula. J Vasc Surg. 2008;47:1330–2.

10. Lowe C, Chan A, Wilde N, Hardy S. Infected endovascular aneu-rysm repair graft complicated by vertebral osteomyelitis. J VascSurg. 2012;56:826–8.

11. Blanch M, Berjón J, Vila R, et al. The management of aortic stent-graft infection: endograft removal versus conservative treatment.Ann Vasc Surg. 2010;24:554.e1–5.

12. Eckroth-Bernard K, Garvin R, Ryer E. Current status ofendovascular devices to treat abdominal aortic aneurysms.Biomed Eng Comput Biol. 2013;5:25–32.

13. Fiorani P, Speziale F, Calisti A, et al. Endovascular graft infection:preliminary results of an international enquiry. J Endovasc Ther.2003;10:919–27.

14. Sharif M, Bernard L, Lau L, et al. Prosthetic stent graft infectionafter endovascular abdominal aortic aneurysm repair. J Vasc Surg.2007;46:442–8.

15. Vogel TR, Symons R, Flum DR. The incidence and factors associ-ated with graft infection after aortic aneurysm repair. J Vasc Surg.2008;47:264–9.

16. McHenry MC, Rehm SJ, Krajewski LP, Duchesneau PM, LevinHS, Steinmuller DR.Vertebral osteomyelitis and aortic lesions: casereport and review. Rev Infect Dis. 1991;13:1184–94.

17. Watura K, Katsimihas M, Williams M. Streptococcus pneumoniaemycotic aneurysm with contiguous vertebral discitis treated byendovascular aortic repair and antibiotics. BMJ Case Rep. 2013;https://doi.org/10.1136/bcr-2012-008499.

18. Stokes W, Janvier J, Vaughan S. Chronic Q fever in Alberta: a caseof Coxiella burnetii mycotic aneurysm and concomitant vertebralosteomyelitis. Can J Infect Dis Med Microbiol. 2016; https://doi.org/10.1155/2016/7456157.

19. Oichi T, Sasaki S, Tajiri Y. Spondylodiscitis concurrent with infec-tious aortic aneurysm caused by Candida tropicalis: a case report. JOrthop Surg (Hong Kong). 2015;23:251–4.

20. Faccenna F, Alunno A, Castiglione A, Carnevalini M, Venosi S,Gossetti B. Large aortic pseudoaneurysm and subsequentspondylodiscitis as a complication of endovascular treatment ofiliac arteries. Thorac Cardiovasc Surg. 2013;61:606–9.

21. Chang CY, Simeone FJ, Nelson SB, Taneja AK, Huang AJ. Isbiopsying the paravertebral soft tissue as effective as biopsyingthe disk or vertebral endplate? 10-year retrospective review ofCT-guided biopsy of diskitis-osteomyelitis. AJR Am JRoentgenol. 2015;205:123–9.

22. De Lucas EM, González Mandly A, Gutiérrez A, et al. CT-guidedfine-needle aspiration in vertebral osteomyelitis: true usefulness of acommon practice. Clin Rheumatol. 2009;28:315–20.

23. Ahuja N, Sharma H. The effectiveness of computed tomography-guided biopsy for the diagnosis of spondylodiscitis: an analysis ofvariables affecting the outcome. Eur Rev Med Pharmacol Sci.2017;21:2021–6.

24. Ledermann HP, Schweitzer ME, Morrison WB, Carrino JA. MRimaging findings in spinal infections: rules or myths? Radiology.2003;228:506–14.

25. Lu W, Pauly KB, Gold GE, Pauly JM, Hargreaves BA. SEMAC:slice encoding for metal artifact correction in MRI. Magn ResonMed. 2009;62:66–76.

26. Hargreaves BA, Worters PW, Pauly KB, Pauly JM, Koch KM,Gold GE. Metal-induced artifacts in MRI. AJR Am J Roentgenol.2011;197:547–55.

27. Haulon S, Lions C, McFadden EP, et al. Prospective evaluation ofmagnetic resonance imaging after endovascular treatment ofinfrarenal aortic aneurysms. Eur J Vasc Endovasc Surg. 2001;22:62–9.

28. Hiramoto JS, Reilly LM, Schneider DB, Skorobogaty H, Rapp J,Chuter TA. The effect of magnetic resonance imaging on stainless-steel Z-stent-based abdominal aortic prosthesis. J Vasc Surg.2007;45:472–4.

29. Talbot BS, Weinberg EP. MR imaging with metal-suppression se-quences for evaluation of total joint arthroplasty. Radiographics.2016;36:209–25.

30. Koch KM, Lorbiecki JE, Hinks SR, King KF. A multi spectralthree-dimensional acquisition technique for imaging near metal im-plants. Magn Reson Med. 2009;61:381–90.

31. Kaushik SS, Karr R, Runquist M, et al. Quantifying metal-inducedsusceptibility artifacts of the instrumented spine at 1.5T using fast-spin echo and 3D-multispectral MRI. J Magn Reson Imaging.2017;45:51–8.

32. Song KD, Yoon YC, Park J. Reducing metallic artefacts in post-operative spinal imaging: slice encoding for metal artefact correc-tion with dual-source parallel radiofrequency excitation MRI at3.0 T. Br J Radiol. 2013;86(1027):20120524.

33. Gratz S, Dörner J, Fischer U, et al. 18F-FDG hybrid PET in patientswith suspected spondylitis. Eur J Nucl MedMol Imaging. 2002;29:516–24.

34. Schmitz A, Risse JH, Grünwald F, Gassel F, Biersack HJ, SchmittO. Fluorine-18 fluorodeoxyglucose positron emission tomographyfindings in spondylodiscitis: preliminary results. Eur Spine J.2001;10:534–9.

35. Stumpe KD, Zanetti M, Weishaupt D, Hodler J, Boos N, VonSchulthess GK. FDG positron emission tomography for differenti-ation of degenerative and infectious endplate abnormalities in thelumbar spine detected on MR imaging. AJR Am J Roentgenol.2002;179:1151–7.

36. Fuster D, Tomás X, Mayoral M, et al. Prospective comparison ofwhole-body (18)F-FDG PET/CT and MRI of the spine in the diag-nosis of haematogenous spondylodiscitis. Eur J Nucl Med MolImaging. 2015;42:264–71.

37. Smids C, Kouijzer IJ, Vos FJ, et al. A comparison of the diagnosticvalue ofMRI and 18F-FDG-PET/CT in suspected spondylodiscitis.Infection. 2017;45:41–9.

38. Tamm AS, Abele JT. Bone and gallium single-photon emissioncomputed tomography-computed tomography is equivalent tomag-netic resonance imaging in the diagnosis of infectiousspondylodiscitis: a retrospective study. Can Assoc Radiol J.2017;68:41–6.

39. Raghavan M, Lazzeri E, Palestro CJ. Imaging of spondylodiscitis.Semin Nucl Med. 2018;48:131–47.

Skeletal Radiol (2018) 47:1357–1369 1369